A Novel Two-Phase Method for the Classification of Incomplete Data

The issue of incomplete data exists across the entire field of data mining. In this paper, a novel two-phase method is developed to deal with the challenge of incomplete data on classification problems. In phase I, the dataset is divided into disjoint subsets based on the attributes with missing values. In phase II, each subset is used to train appropriate classification algorithms respectively in parallel. Experimental results show that the proposed scheme works favorably compared to other techniques on both synthesized and real data sets.